-
公开(公告)号：US20120072215A1
公开(公告)日：2012-03-22
申请号：US12886568
申请日：2010-09-21
IPC分类号： G10L15/14
CPC分类号： G06K9/6296 , G06N3/0454 , G06N3/084 , G10L15/14
摘要： A method is disclosed herein that include an act of causing a processor to access a deep-structured model retained in a computer-readable medium, wherein the deep-structured model comprises a plurality of layers with weights assigned thereto, transition probabilities between states, and language model scores. The method can further include the act of jointly substantially optimizing the weights, the transition probabilities, and the language model scores of the deep-structured model using the optimization criterion based on a sequence rather than a set of unrelated frames.
摘要翻译： 本文公开了一种方法，其包括使处理器访问保留在计算机可读介质中的深层结构模型的动作，其中深层结构化模型包括赋予权重的多个层，状态之间的转移概率和 语言模型得分。 该方法还可以包括使用基于序列而不是一组不相关帧的优化准则共同基本优化深层结构模型的权重，转移概率和语言模型分数的动作。
-
2.发明申请SYSTEM AND METHOD FOR APPLYING A CONVOLUTIONAL NEURAL NETWORK TO SPEECH RECOGNITION 有权用于将语音神经网络应用于语音识别的系统和方法公开(公告)号：US20150255062A1
公开(公告)日：2015-09-10
申请号：US14720891
申请日：2015-05-25
申请人： Gerald Bradley Penn , Hui Jiang , Ossama Abdelhamid Mohamed Abdelhamid , Abdel-rahman Samir Abdel-rahman MOHAMED
发明人： Gerald Bradley Penn , Hui Jiang , Ossama Abdelhamid Mohamed Abdelhamid , Abdel-rahman Samir Abdel-rahman MOHAMED
IPC分类号： G10L15/16
摘要： A system and method for applying a convolutional neural network (CNN) to speech recognition. The CNN may provide input to a hidden Markov model and has at least one pair of a convolution layer and a pooling layer. The CNN operates along the frequency axis. The CNN has units that operate upon one or more local frequency bands of an acoustic signal. The CNN mitigates acoustic variation.
摘要翻译： 一种将卷积神经网络（CNN）应用于语音识别的系统和方法。 CNN可以向隐马尔可夫模型提供输入，并且具有至少一对卷积层和汇集层。 CNN沿着频率轴运行。 CNN具有对声信号的一个或多个本地频带进行操作的单元。 CNN减轻声学变化。
-
3.发明授权System and method for applying a convolutional neural network to speech recognition 有权将卷积神经网络应用于语音识别的系统和方法公开(公告)号：US09190053B2
公开(公告)日：2015-11-17
申请号：US13849793
申请日：2013-03-25
申请人： Gerald Bradley Penn , Hui Jiang , Ossama Abdelhamid Mohamed Abdelhamid , Abdel-rahman Samir Abdel-rahman Mohamed
发明人： Gerald Bradley Penn , Hui Jiang , Ossama Abdelhamid Mohamed Abdelhamid , Abdel-rahman Samir Abdel-rahman Mohamed
摘要： A system and method for applying a convolutional neural network (CNN) to speech recognition. The CNN may provide input to a hidden Markov model and has at least one pair of a convolution layer and a pooling layer. The CNN operates along the frequency axis. The CNN has units that operate upon one or more local frequency bands of an acoustic signal. The CNN mitigates acoustic variation.
摘要翻译： 一种将卷积神经网络（CNN）应用于语音识别的系统和方法。 CNN可以向隐马尔可夫模型提供输入，并且具有至少一对卷积层和汇集层。 CNN沿着频率轴运行。 CNN具有对声信号的一个或多个本地频带进行操作的单元。 CNN减轻声学变化。
-
公开(公告)号：US09031844B2
公开(公告)日：2015-05-12
申请号：US12886568
申请日：2010-09-21
CPC分类号： G06K9/6296 , G06N3/0454 , G06N3/084 , G10L15/14
摘要： A method includes an act of causing a processor to access a deep-structured model retained in a computer-readable medium, the deep-structured model includes a plurality of layers with respective weights assigned to the plurality of layers, transition probabilities between states, and language model scores. The method further includes the act of jointly substantially optimizing the weights, the transition probabilities, and the language model scores of the deep-structured model using the optimization criterion based on a sequence rather than a set of unrelated frames.
摘要翻译： 一种方法包括使处理器访问保存在计算机可读介质中的深层结构模型的行为，所述深层结构模型包括分配给所述多个层的各个权重的多个层，状态之间的转移概率和 语言模型得分。 该方法还包括使用基于序列而不是一组不相关帧的优化准则来共同基本优化深层结构模型的权重，转移概率和语言模型分数的动作。
-
5.发明申请SYSTEM AND METHOD FOR APPLYING A CONVOLUTIONAL NEURAL NETWORK TO SPEECH RECOGNITION 有权用于将语音神经网络应用于语音识别的系统和方法公开(公告)号：US20140288928A1
公开(公告)日：2014-09-25
申请号：US13849793
申请日：2013-03-25
申请人： Gerald Bradley Penn , Hui Jiang , Ossama Abdelhamid Mohamed Abdelhamid , Abdel-rahman Samir Abdel-rahman Mohamed
发明人： Gerald Bradley Penn , Hui Jiang , Ossama Abdelhamid Mohamed Abdelhamid , Abdel-rahman Samir Abdel-rahman Mohamed
IPC分类号： G10L15/16
摘要： A system and method for applying a convolutional neural network (CNN) to speech recognition. The CNN may provide input to a hidden Markov model and has at least one pair of a convolution layer and a pooling layer. The CNN operates along the frequency axis. The CNN has units that operate upon one or more local frequency bands of an acoustic signal. The CNN mitigates acoustic variation.
摘要翻译： 一种将卷积神经网络（CNN）应用于语音识别的系统和方法。 CNN可以向隐马尔可夫模型提供输入，并且具有至少一对卷积层和汇集层。 CNN沿着频率轴运行。 CNN具有对声信号的一个或多个本地频带进行操作的单元。 CNN减轻声学变化。
-
-
-
-
搜索结果
5 条记录 (耗时 0.013 秒)
